Method of Fabricating a Mos Device With Non-Sio2 Gate Dielectric
A polycrystalline silicon layer is deposited on a gate dielectric and then a portion thereof is re-oxidized so as to form a thin layer of oxide between the poly-Si layer and the underlying gate dielectric. Subsequently, the poly-Si layer is converted to a fully-silicided form so as to produce a FUSI gate. The gate dielectric can be a high-k material, for example a Hf-containing material, or SION, or another non-SiO2 dielectric. The barrier oxide layer is preferably less than 1 nm thick.
Latest Freescale Semiconductor, Inc. Patents:
- AIR CAVITY PACKAGES AND METHODS FOR THE PRODUCTION THEREOF
- METHODS AND SYSTEMS FOR ELECTRICALLY CALIBRATING TRANSDUCERS
- SINTERED MULTILAYER HEAT SINKS FOR MICROELECTRONIC PACKAGES AND METHODS FOR THE PRODUCTION THEREOF
- CONTROLLED PULSE GENERATION METHODS AND APPARATUSES FOR EVALUATING STICTION IN MICROELECTROMECHANICAL SYSTEMS DEVICES
- SYSTEMS AND METHODS FOR CREATING BLOCK CONSTRAINTS IN INTEGRATED CIRCUIT DESIGNS
The present invention relates to the field of integrated circuit device manufacture. More particularly, the present invention relates to the manufacture of metal-oxide-silicon (MOS) devices and the like, notably such devices employing high dielectric constant (high-k) gate dielectric or, more generally, non-SiO2 gate dielectric, that is, gate dielectric materials which do not behave as classic SiO2 dielectric does, even if based on SiO2 (thus including doped SiO2 dielectrics).
The continuing drive for greater scaling of integrated circuit devices has lead to a desire to replace the SiO2 that is traditionally used as gate dielectric in MOS devices. The Semiconductor Industry Association's (SIA's) International Technology Roadmap for Semiconductors (ITRS) 2004 specifies that MOSFET devices used in CMOS will require a gate-oxide equivalent thickness (EOT) less than 1.4 or 1.5 nm when these devices are scaled so as to have gate lengths below 65 nm, dropping to 0.8 nm or less by 2010. This would be very hard, if not impossible, to achieve using SiO2 gate dielectric because it would require the use of a SiO2 gate dielectric layer only a few atoms thick. Moreover, because of quantum mechanical direct tunneling, leakage current increases as the thickness of the SiO2 gate dielectric decreases.
Accordingly, high-k dielectric materials have been attracting attention for use as gate dielectric materials, notably in applications where low gate leakage current is desired. High-k dielectric materials are those having a dielectric constant, k greater than that of silicon nitride, i.e. k>7. Some high-k dielectric materials under consideration include oxides and silicates of metals such as tantalum (Ta), zirconium (Zr) and hafnium (Hf). A number of existing proposals involve use of gate dielectric materials containing Hf, notably, HfO2, HfO2/SiN, HfSiON and HfAlO.
Difficulties arise when it is desired to use high-k dielectric materials, instead of SiO2, as gate dielectric. In particular, SiO2 is compatible both with the Si substrate and with existing fabrication processes. In order to be able to use a high-k dielectric material, or indeed other non-SiO2 dielectric materials, as the gate dielectric, it is necessary to overcome potential problems due to chemical bonding effects and difficulties arising from the desirability of integrating the new material into existing fabrication processes.
For example, in recent years MOSFET gate electrodes have habitually been made of polycrystalline silicon (poly-Si). Various problems arise when high-k dielectric materials (or other non-SiO2 dielectrics) are used as gate dielectric materials with poly-Si gates.
In particular, when using poly-Si on non-SiO2 gate dielectric, carrier mobility in the MOSFET device can be low. Moreover, the threshold voltage, Vt, is different from that observed when SiO2 dielectric is used with poly-Si gate electrodes: typically Vt is −0.3 V for poly-Si on SiO2 whereas it is typically in the range −0.6 to −0.9 V for poly-Si on high-k gate dielectric. Furthermore, when poly-Si is used with high-k gate dielectric it is generally observed that the threshold voltage, Vt, for PMOS devices is too high, (because of Fermi-level pinning of the gate work function).
One approach that has been tried in view of the above-mentioned problems is to convert the poly-Si gate electrode material to silicide, for example NiSi or CoSi, thus forming a fully-silicided (or “FUSI”) gate electrode. This can be considered to be a metal-like gate electrode. This approach has the advantage of avoiding the poly-depletion effect that is observed when using poly-Si gate electrodes, thus reducing the equivalent oxide thickness (EOT) at the gate. Certain proposals have used Hf-based materials (notably HfSiON, HfO2 and HfOxNy) as a high-k gate dielectric material in association with FUSI gate electrodes.
However, when FUSI gate electrodes are used with high-k gate dielectric materials the following problem arises: there is insufficient PMOS/NMOS Vt modulation (asymmetric Vt and higher-than-target Vt values).
In view of the above problems, various researchers have concluded that it is preferable to use SiON rather than high-k dielectric materials when selecting a material for use as the gate dielectric for a CMOS device having a FUSI-gate. (SiON was already being considered as a replacement for SiO2 gate dielectrics as an intermediate step on the way to changing to high-k dielectric materials, in view of the fact that a SiON film has a dielectric constant of about 4 to 8, depending upon the proportions of O and N therein.) However, it is generally found that the leakage current is undesirably high in devices using FUSI-gate electrodes on SiON gate dielectrics.
An alternative approach that has been tried in order to overcome the problems involved in using high-k gate dielectric with poly-Si gate electrodes is to modify the interface between the high-k gate dielectric and the poly-Si gate electrode, for example by adding barrier layers (or “capping layers”) between the gate dielectric and the poly-Si gate electrode. However, this approach has not been particularly successful—even when a variety of different materials and deposition techniques were used to form the barrier layers on the high-k material, there was no reduction, or inadequate reduction, in the observed Vt shift when the gate stack structure was subsequently completed by a poly-Si gate electrode, and/or carrier mobility was degraded. In some cases the capping technique was unsuitable for making NMOS devices. Also, introduction of capping layers causes undesirable increase in the EOT.
The inadequacy of capping for tackling Vt shift can be seen from
The first comparative example is a PMOSFET structure using a SiO2 gate dielectric 2.5 nm thick; the results for this first comparative example are indicated using hollow squares in
As seen from
Poor results are seen for capped high-k dielectric materials used with poly-Si gate electrodes even when the high-k dielectric is capped with SiO2.
The present inventor has postulated that the reason for the disappointing results that are observed in gate stacks using poly-Si gate electrodes and high-k dielectrics with caps is that the capping layers are formed on the gate dielectric layer before the poly-Si layer has been deposited. In other words, at the time when the capping layers are formed there is not yet an interface between the high-k dielectric material and the poly-Si.
The present invention provides a new method of fabricating MOS devices comprising FUSI gates, as defined in the appended claims.
The present invention further provides a MOS device comprising FUSI gates, as defined in the appended claims.
In the preferred embodiments of the present invention, a poly-Si layer is formed on a gate dielectric layer and then a layer of oxide is produced between the poly-Si and the gate dielectric, notably by a lateral re-oxidation process.
The formation of the thin oxide layer after the poly-Si has already been deposited on the gate dielectric releases Fermi-level pinning of the gate work function, enabling lower threshold voltages (Vt) to be obtained for smaller gates. As a result, a desired work function modulation can be achieved.
Because the lateral reoxidation process has been found to release Fermi-level pinning and reduce Vt shift it is believed that the thin layer of oxide formed by this process is generated by oxidation of the poly-Si.
Although the reoxidation process leads to an increase in the EOT at the gate, a CETinv (Capacitance equivalent thickness in inversion) reduction can be obtained by use of a FUSI gate—in other words, by converting the poly-Si layer into a fully-silicided form. Thus, the increase in EDT from a physically thicker dielectric is compensated by the absence of polysilicon depletion using a metal-like FUSI gate, so that overall there is little or no increase in CETinv under transistor operation conditions of inversion.
In certain of the preferred embodiments of the invention, the reoxidation process forms a particularly thin oxide layer between the gate electrode and gate dielectric, notably a layer having a thickness less than 10 angstroms (<1 nm).
In certain of the preferred embodiments of the invention, the gate dielectric is a high-k dielectric material, for example a Hf-containing material.
An advantage of fabricating MOS devices using FUSI gate electrodes on high-k gate dielectrics is that this enables further scaling of ULSI devices to be achieved whilst still using existing fabrication equipment. This avoids the need to introduce dual metal gate electrode devices, which could require significant changes in the machinery used in ULSI device fabrication processes.
The above and other features and advantages of the present invention will become apparent from the following description of a preferred embodiment thereof, given by way of example, and illustrated in the accompanying drawings, in which:
The preferred embodiment of the present invention will now be described with reference to
As shown in
If desired, a capping layer (not shown) can be formed on the gate dielectric 10 before the subsequent processing. This will be appropriate, notably, when the high-k material forming the gate dielectric layer 10 acts as a barrier to penetration of oxygen (e.g. because of a post-deposition treatment such as plasma nitridation). According to the present invention it is desirable that the top few angstroms of the gate dielectric allow oxygen to pass therethrough (so as to oxidize the bottom few angstroms of the polysilicon gate and form the thin ‘special’ layer that releases the Fermi level pinning, as explained further below). Thus, in a case where there is poor oxygen diffusion through the high-k material forming the gate dielectric layer 10 it can be advantageous to deposit a cap layer formed of a small amount of HfO2, or another excellent oxygen conductor (this will help achieve reoxidation at a lower temperature or more rapidly).
A poly-Si layer 20 is formed by any convenient process, for example: chemical vapor deposition (CVD), plasma-assisted CVD, sputtering, etc., on the gate dielectric 10 (or on its capping layer, if applicable). At this stage, the poly-Si layer 20 will typically have a thickness of 10 nm to 200 nm. The technique used for depositing the poly-Si layer is not critical to the success of the invention. However, from a practical viewpoint, it is preferred to use a deposition technique that results in formation of a layer of fine-grained or columnar crystalline poly-Si (so as to oxidize rapidly during the lateral reoxidation step discussed below).
The poly-Si layer 20 is patterned in accordance with the desired dimensions of the gate electrode, to produce a structure as illustrated schematically in
Any convenient process can be used for patterning the poly-Si layer 20, for example, in some cases a standard dry etch process can be used. If the gate dielectric layer 10 is formed of a high-k material then the standard dry etch process can be modified in a known manner so as to ensure removal of the high-k dielectric from areas exposed during etching of the gate electrode, without creation of a recess in the underlying Si substrate.
After the gate electrode patterning has been performed, a controlled lateral oxidation process is performed in order to produce a barrier layer 30 of silicon dioxide (SiO2) at the interface between the poly-Si material 20 and the gate dielectric 10, as shown in
A lateral reoxidation process can successfully produce a thin SiO2 layer 30 between the HfO2 gate dielectric 10 and the poly-Si layer 20. This is illustrated in
In the example illustrated by
To produce a SiO2 layer of a sufficiently low thickness for practical application, grown by lateral oxidation, between the polysilicon and the high-k layers, it is advisable to use a low temperature 500-700° C. oxidation in a batch-wafer furnace or a 600-800° C. rapid thermal oxidation in a single wafer tool. Since it is difficult to estimate the thickness of laterally-grown SiO2 on device wafers, the oxidation condition is verified on bare Si monitor wafers that are oxidized using the same conditions as the device wafers. The thickness of the laterally-grown SiO2 will be much lesser than that grown on a bare SiO2 wafer.
Further, the oxidation condition will depend on the gate dielectric material, particularly the oxygen diffusion properties of the material. Rapid lateral diffusion of oxygen through the dielectric layer will allow a shorter time and/or lower temperature oxidation sequence. HfO2 allows fast diffusion of the oxygen laterally, while Hf-silicate allows slower diffusion. HfSiON will likely allow very slow oxygen diffusion, thereby requiring stronger (higher temperature or longer time) oxidation conditions. An oxidation condition that produces a target SiO2 layer of 1-2 nm on a bare silicon wafer is optimal for growing a thin SiO2 layer between the gate dielectric layer 10 and the poly-Si layer 20 by lateral oxidation.
According to the preferred embodiment of the present invention, the thickness of the oxide barrier layer 30 should be related to the gate length. The extent of lateral oxidation across a patterned poly-Si line has a strong dependence on the oxidation temperature, partial pressure, time of oxidation, the permeability of oxygen through the dielectric, and a weak dependence on the oxidation rate of the polysilicon grain. The relationship is mathematically complex, but it has been found that the oxidation thickness saturates at a value determined by the diffusion limitation.
More particularly, it has been found to be advantageous to set the thickness of the oxide layer 30 to less than 10 angstroms (1 nm), for example to a value of 6 angstroms.
In principle, a minimum SiO2 layer that is 1 monolayer or ˜0.3 nm thick will be sufficient to release the Fermi-level pinning and bring the Vt values for poly-Si on non-SiO2 dielectric close to that of poly-Si on SiO2. However, in practice, it may be difficult to ensure the uniform growth of a 0.3 nm layer of SiO2 by lateral reoxidation. As a practical matter, it is simple to set the oxidation conditions with the aim of producing a 0.5 nm+/−0.1 nm SiO2 layer. A SiO2 layer of this thickness gives the benefit of Vt-shift reduction while only causing a limited increase in the EOT. This increase in EOT can be compensated by converting the poly-Si layer 20 to a fully-silicided form, making it possible to take advantage of the poly depletion gain (˜0.4 nm) obtained from use of the metal-like FUSI gate. With this choice of thickness and process conditions, the benefit of Vt reduction, or Vt matching with SiO2 and poly depletion, can be achieved.
The subsequent steps in the method according to the preferred embodiment of the present invention (illustrated in
The present inventor has found that the devices fabricated using the method according to the preferred embodiment of the invention have a reduced Vt shift as the gate length is scaled down. This improvement in device performance is due to the thin barrier layer of oxide 30 that is produced by the above-described lateral reoxidation process.
The improved Vt shift can be seen from a consideration of
In
The first example according to the invention is a PMOSFET structure based on the above-described second comparative example; it uses a HfO2 gate dielectric 3 nm thick (without a SiNx capping layer) and, between the poly-Si gate electrode and the gate dielectric layer, there is an SiO2 barrier layer formed by lateral oxidation under process conditions in a batch furnace so as to give a 2 nm thick SiO2 layer on a bare wafer at 700° C. The exact thickness of this layer on the device wafer is seen by XTEM later. The results for this first example are indicated using hollow circles in
The second example according to the invention is a PMOSFET structure based on the above-described third comparative example; it uses a HfO2 gate dielectric 2.5 nm thick topped by a HfSiOx capping layer 1 nm thick and, between the poly-Si gate electrode and the gate dielectric layer, there is an SiO2 barrier layer formed by lateral oxidation under process conditions in a batch furnace so as to give a 2 nm thick barrier layer 30 at 700° C. The results for this second example are indicated using hollow triangles in
A comparison of the results shown in
A comparison of the results shown in
In
The third example is a PMOSFET structure having, between the HfSiO dielectric-capping layer and the poly-Si gate electrode, a laterally-grown SiO2 layer (grown under rapid thermal oxidation conditions at 800° C. and 1 Torr for 34 seconds—conditions which grow a 1 nm oxide layer on a bare Si wafer). The results for this third example are indicated using hollow upright triangles in
The fourth example is a PMOSFET structure having, between the HfSiO dielectric-capping layer and the poly-Si gate electrode, a laterally-grown SiO2 layer (grown under rapid thermal oxidation conditions at 800° C. and 1 Torr for 68 seconds—conditions which grow a <2 nm thick oxide layer on a bare Si wafer). The results for this fourth example are indicated using hollow circles in
The fifth example is a PMOSFET structure having, between the HfSiO dielectric-capping layer and the poly-Si gate electrode, an oxide layer 30 formed by in-situ steam generation (ISSG) in a cold wall rapid thermal processing chamber, with the wafer held at 800° C. for 26 seconds in N2O at 12.5 Torr pressure (targeted to grow 1 nm nominally on a bare Si wafer). The results for this fifth example are indicated using hollow sideways triangles in
The sixth example according to the invention is a PMOSFET structure having, between the HfSiO dielectric-capping layer and the poly-Si gate electrode, an oxide layer 30 formed by ISSG in a cold wall rapid thermal processing chamber, with the wafer held at 800° C. for 52 seconds in N2O at 12.5 Torr. The results for this sixth example are indicated using hollow diamonds in
The above results demonstrate that the formation of a thin oxide barrier layer between the high-k dielectric and the poly-Si electrode is effective to release Fermi-level pinning and provide an acceptable work function modulation. Moreover, poly depletion is improved. However, even when the thickness of the oxide barrier layer is minimized the creation of this oxide barrier layer leads to an increase of around 0.4 nm in the EOT at the gate. Accordingly, it is desirable to adopt some measure to counteract this increase in EOT.
It is known that devices using FUSI gates on high-k dielectrics or SiON have a significantly lower CETinv than comparable devices using poly-Si gates. In particular, a reduction in CETinv of around 0.3 nm can be seen in devices having FUSI gate electrodes compared to devices having poly-Si gate electrodes.
According to the preferred embodiments of the present invention, after the oxide barrier layer 30 has been formed, the poly-Si layer 20 is converted to a fully-silicided form to produce a FUSI-gate electrode, thereby reducing the EOT of the device.
Returning to
The resulting gate stack structure is sintered typically at temperatures between 300-500° C. whereby the metal species diffuses into the poly-Si, reacting with it fully (down to the interface with the barrier oxide 30). This produces a fully-silicided layer 50, as indicated in
The layer 50 can have any desired thickness compatible with the poly-Si process. The thickness of the metallic layer 40 (e.g. Ni) is adjusted in dependence on the thickness of the poly-Si layer 20 so as to give full silicidation of the poly-silicon. The techniques and conditions necessary to produce FUSI gate electrodes are well-known to the skilled person and so no further details will be given here.
It has been found that the method according to the preferred embodiment of the present invention can produce MOS devices that have FUSI gate electrodes and high-k gate dielectric with adequate work function modulation. This makes it feasible to use devices having FUSI electrodes at the 65 nm and 45 nm technology nodes. Thus, it will still be possible to use conventional fabrication equipment even when ULSI device features are scaled down to 45/65 nm. More particularly, known processes and equipment can be used to implement the various steps of the present invention, with only minor adaptations being required in processing conditions for reoxidation or silicidation steps (whereby to ensure compatibility of these processes with the gate dielectric material and whereby to form a thin lateral-oxidation-grown SiO2 layer).
Although the present invention has been described above with reference to a particular preferred embodiment, it is to be understood that the invention is not limited by reference to the specific details of this preferred embodiment. More specifically, the person skilled in the art will readily appreciate that, modifications and developments can be made in the preferred embodiment without departing from the scope of the invention as defined in the accompanying claims.
For example, in the above-described preferred embodiment a thin oxide layer is grown between a poly-Si layer formed on a high-k dielectric, notably a Hf-containing high-k dielectric. However, the present invention can be applied to grow an oxide layer between a poly-Si layer formed on other high-k dielectric materials, and on other non-SiO2 dielectric materials, such as a SiON gate dielectric layer.
Furthermore, in the above-described preferred embodiment of the invention, the poly-Si layer was patterned to form the desired shape of the gate electrode before the lateral reoxidation step was performed. However, alternative approaches are possible. For example, a super-thin layer of poly-Si may be deposited over the dielectric layer 10 on the wafer, then fully oxidized, and then thinned down to an allowable thickness by etching.
Moreover, although the present invention has been described above in terms of embodiments relating to the fabrication of a gate stack structure for a MOSFET device, the present invention is applicable to the fabrication of other devices, notably devices where threshold voltage control is important but is harder to achieve when the dielectric is changed from SiO2 to a non-SiO2 dielectric (e.g. a high-k dielectric).
Claims
1. A method of fabricating a MOS device, comprising the steps of:
- forming a layer of gate dielectric material on a substrate; and
- forming a layer of polycrystalline silicon on said gate dielectric layer;
- forming a layer of oxide between said polycrystalline silicon layer and said gate dielectric layer after said layer of polycrystalline silicon has been formed on said layer of gate dielectric material; and
- converting said polycrystalline silicon to a fully-silicided form.
2. The MOS device fabrication method of claim 1, wherein said oxide-layer formation step comprises process steps adapted to cause the formation of an oxide layer having a thickness of less than 1 nm.
3. The MOS device fabrication method of claim 1, wherein said gate dielectric material is a high-k dielectric material.
4. The MOS device fabrication method of claim 3, wherein said gate dielectric material is a Hf-containing material.
5. The MOS device fabrication method of claim 1, wherein said gate dielectric material is SiON.
6. A method of claim 1, further comprising forming a gate dielectric and a FUSI gate electrode, wherein there is a layer of oxide between the gate electrode and the gate dielectric.
7. The method according to claim 6, wherein the oxide layer between the gate electrode and the gate dielectric is less than 1 nm thick.
8. A method according to claim 6, wherein the gate dielectric comprises a layer of high-k dielectric material.
9. A method according to claim 6, wherein the gate dielectric comprises a layer of SiON.
10. The method according to claim 2, wherein said gate dielectric material is a high-k dielectric material.
11. The method according to claim 2, wherein said gate dielectric material is SiON.
12. The method according to claim 2, further comprising forming a gate dielectric and a FUSI gate electrode, wherein there is a layer of oxide between the gate electrode and the gate dielectric.
13. The method according to claim 3, further comprising forming a gate dielectric and a FUSI gate electrode, wherein there is a layer of oxide between the gate electrode and the gate dielectric.
14. The method according to claim 4, further comprising forming a gate dielectric and a FUSI gate electrode, wherein there is a layer of oxide between the gate electrode and the gate dielectric.
15. The method according to claim 5, further comprising forming a gate dielectric and a FUSI gate electrode, wherein there is a layer of oxide between the gate electrode and the gate dielectric.
16. The method according to claim 7, wherein the gate dielectric comprises a layer of high-k dielectric material.
17. The method according to claim 7, wherein the gate dielectric comprises a layer of SiON.
Type: Application
Filed: Apr 21, 2005
Publication Date: Aug 14, 2008
Patent Grant number: 7820538
Applicant: Freescale Semiconductor, Inc. (Austin, TX)
Inventor: Vidya Kaushik (Hoeilaart)
Application Number: 11/911,931
International Classification: H01L 21/283 (20060101);